2.3. Identification of Putative Gene Functions Regulated by AfeR
All Type I PBSs were selected as possible AfeR binding sites. PBSs belonging to the Type II set were selected only if their location almost overlapped or overlapped with a consensus sigma 70 (see Experimental Section). All PBSs belonging to the Type III set were discarded as probable binding sites due to their location. Applying this filter yielded a total of 62 PBSs (
Table S1). Moreover, to improve the reliability of the PBS assignment, sequences of all selected PBSs plus 80 bp flanking each side were used to screen for common sequence motifs using the MEME motif discovery tool [
17]. MEME searches for common motifs in a set of nucleotide or amino acid sequences. We applied the MEME tool without using the
afeI upstream palindromic sequence as an input for this search. The results show that the most frequent and highest scoring motif was an 18-mer that corresponds to the MP based models present in each input sequence. The MEME program found the 18 bp motif in most of the MP-containing input sequences. On the other hand, the SP based PBSs were not detected as a MEME common motif. Interestingly, some of the SP containing input sequences possessed the MEME 18 bp motif near or overlapping the corresponding SP sequence. These new motifs were not detected as MP-based PBSs in the previous HMM search. Nevertheless, the MEME tool mainly confirmed the results obtained with the HMMs. Then, the gene functions associated with all these identified PBs were noted (
Table S1). The identified gene functions whose expression could be controlled by AfeR were diverse, and there was not a clear biasing towards a particular functional category.
Selected PBSs that also shared the MEME 18 bp common motif, including the new ones located in the SP-containing inputs, were aligned and a consensus sequence was generated using MultAlin [
18] (
Figure 3). The output from MEME provides a position-specific scoring matrix (PSSM) for the predicted motif. The PSSM generated for the 18 bp MEM motif was used as an input for the MAST program (Motif Alignment and Search Tool), which was applied to a database consisting of all intergenic regions in the genome [
19]. Eight motifs with e-values lower than 10 were found where the lowest e-value corresponded to the
afeI palindrome. Two out of eight motifs were located in intergenic regions between converging ORFs. Two previously identified motifs were those located upstream the
afeR gene and AFE_2100 (annotated as a virulence-associated protein). One new motif was found in the intergenic region upstream the AFE_1504 ORF (annotated as “nifW protein”). Also a new motif was located between the diverging ORFs AFE_2800 and AFE_2799, which are a glyoxalase family protein and an Ada family-DNA-3-methyladenine glycosylase II transcriptional regulator, respectively. The positions of PBSs that where found using the MEME algorithm are summarized in
Figure 4.
To evaluate the effectiveness of an alternative lux-box consensus-based search, the upstream 250 nucleotides from the four most identical sequences to the product of afeI in the genbank database (luxI-like genes from Burkholderia pseudomallei K96243, Burkholderia cepacia AMMD, Burkholderia thailandensis E264, Burkholderia ambifaria MC40-6) were used as input for the MEME discovery tool (the afeI palindrome containing upstream nucleotides was also used). In all four sequences a 18 bp motif that corresponded to a palindromic sequence similar to the afe-box was found (Results not shown). The PSSM generated was put into the MAST tool and the intergenic region database from At. ferrooxidans was screened for common motifs. The result from this procedure showed that only the afeI 18-bp motif was found. On the contrary, when the same database was screened using the PSSM generated from the MEME analysis of the MP-based PBSs, the number of hits was broader (see results). This suggests that using lux-box like sequences for the generation of HMMs in the At. ferrooxidans search would have yielded fewer hits, and hence a narrower search.
2.4. The At. ferrooxidans Putative Quorum Sensing Regulon
Functions regulated by QS in At. ferrooxidans are unknown. Here, we report results obtained from a bioinformatic approach to identify possible binding sites of the AfeR protein in the At. ferrooxidans ATCC 23270 genome sequence with the aim of defining a putative QS regulon for this bacterium.
HMMs were constructed based on alignments of hypothetical sequences that took the palindromic structure located upstream of the
afeI gene as a template. Alignments of known
lux-box type sequences to generate the HMMs were not performed because of the little overall conservation of these sequences [
20]. The HMMs were designed to conserve palindromic zones and vary non-palindromic zones. This design allowed the retention of the modular nature of palindrome organization as opposed to sequence conservation only. Modularity may have a role in the mechanism of transcriptional regulation at the
afe-box because different modes of binding are expected depending on the organization of sequence modules, as it has been experimentally tested for the
fur binding site [
21]. The fact that the canonical
lux-box is only a fraction of a larger structure suggests that non-AfeR transcriptional regulators might also bind to the
afeI operator.
The search yielded a total of 265 PBSs. SP model-based PBSs were preferentially located on intergenic regions whereas MP model-based PBSs were not (
Figure 2). This can be interpreted as a “preference” of the SP model-based PBSs towards intergenic regions, hence raising the possibility that SP PBSs are true
cis-acting regulatory motifs. Also, it is possible that this preference is due to the overlapping of the -35 element of a sigma 70 promoter from
afeI with the modeled SP structure, thus implying that at least some of the SP hits may actually be -35 elements. Supporting the hypothesis mentioned above, the Bprom predictor found some Type I SP hits overlapping -35 elements but it also found hits overlapping -10 elements and hits with no overlapping at all. On the other hand, only MP model-based PBSs were found as common motifs when all PBSs were searched with the MEME tool, making MP PBSs more likely to be the correct regulatory motif for AfeR. Thus, as SP and MP based models did not show significant overlapping in the HMM search, the determinants for the AfeR interaction with DNA are likely to be enclosed within the central 18–20 nucleotides of the large 30 bp palindrome and that the SP PBSs may be binding sites for a distinct regulator. It is possible then that SP based PBSs may be part of a distinct regulon that overlaps the AfeR regulon in the transcriptional regulation of
afeI. There are examples of multiple regulatory
cis-acting elements upstream of various QS-controlled genes, including the regulatory zones of I and R genes [
22,
23]. These regulatory sequences can overlap
lux-box like elements as is the case of the binding site for the regulator RsaL, upstream of
lasI from
P. aeruginosa [
24] or the MetR (LysR-type transcriptional regulator) and CRP (catabolite repressor protein) binding sites upstream of
luxI gene from
V. fischeri [
23].
Among the ORFs presented in the PBS table (
Supplementary Table 1), five of them had two PBSs and one had three PBSs upstream of their predicted ATG start codon. Hypothetical protein AFE_1942, the diverging genes AFE_0999 (conserved domain protein) and AFE_0998 (putative transcriptional regulator), and the ErfK-YbiS-YcfS-YnhG family protein AFE_0569 had a SP model-based PBSs and a predicted MEME 18-bp motif. In the case of AFE_0569, the sequences are overlapped (
Supplementary Table 1). AFE_1354 (a group 1 glycosyl transferase family protein) has 2 MP-based PBSs upstream from its start codon. Hypothetical protein AFE_1055 has two SP-based PBSs and a MEME 18-bp motif overlapping one of these. Examples exist where more than one
lux-box like sequence is found upstream of QS-controlled genes. This is also the case for the
lasB,
hcn and
pqsA genes from
P. aeruginosa [
25–
27].
The type of regulation exerted over the AfeR targets remains unknown. AfeR can activate transcription from the
afeI promoter in the presence of AHL [
13], but the binding dependence on AHL remains to be determined. The ORFs found to be putatively regulated by the AfeR protein in this study have their respective PBSs located in various positions relative to predicted σ
70 promoters. Some of them overlap -35 elements, others overlap -10 elements and others do not overlap either. Previously described LuxR family activators of transcription like TraR, LasR and LuxR itself generally bind a
lux-box like element centered in position -43 (overlapping the -35 element of the σ
70 promoter) in the presence of AHL. Some transcriptional regulators of the LuxR family bind their target sequences in the absence of and self-dissociate in the presence of AHL, as is the case of EsaR of
Pantoea stewartii or ExpR
Ecc of
Erwinia carotovora [
28,
29]. Both mentioned regulators are repressors of transcription. In this case, the location of the
lux-box like sequence overlaps the -10 element of the σ
70 promoter. The mentioned regulators can also function as activators of transcription in artificial genetic constructs where the position of the
cis-acting element is changed to a -35 element overlap (maintaining the dependence on absence of AHL substrate) [
30]. This shows that the ability to interact with RNA polymerase and activate transcription remains intact, and that activation rather than repression of other genes in its native context can occur depending on the
lux-box like element position. Moreover, it has been demonstrated that genuine LasR binding sites are located as far as 383 bp from the translation start codon (
phzA gene) [
31], thus making our screening somewhat conservative.
None of the PBSs found had similar palindromic complexity to neither the
afeI upstream sequence nor perfect dyad symmetry. Dyad symmetry is not a prerequisite for the
lux-box like elements to interact with their cognate regulator as demonstrated for various QS-controlled genes in
P. aeruginosa [
20].
The gene functions associated to the 18 bp MEME motif have been related to several biological processes (
Figure 4). One of these is QS itself, represented by AfeR and AfeI proteins and possibly the product of the
orf3 gene. The position of the
afe-box located upstream of o
rf3 and
afeR overlaps predicted -35 elements in both cases, suggesting an activator role for AfeR at this locus. Other LuxR-like proteins are also subject of self-regulation, as is the case for the positive feedback loop experienced by PhzR from
Pseudomonas fluorescens [
32] and the negative feedback loop involving EsaR from
Pantoea stewartii [
29].
There were two other transcriptional regulators forming part of the predicted targets: first, a helix-turn-helix transcriptional regulator that also contains a LexA motif and second, a Sigma 54-dependent transcriptional regulator. This is consistent with the fact that AfeR is a global regulator of transcription that controls the expression of many genes indirectly. The S24/LexA-like domain is related to the bacterial SOS response; specifically it catalyzes its own proteolysis separating the DNA-binding domain and the rest of the protein. (CD number: cd06529) The Sigma 54-dependent transcriptional regulator has also an HTH DNA-binding motif and also an AAA-type ATPase domain (CD number: cd00009).
In the same locus as sigma 54-dependent transcriptional regulator, but in the opposite orientation, there is a metallo-β-lactamase protein family gene. Interestingly, there is the second metallo-β-lactamase among the hypothetical regulon. These proteins belong to the same superfamily as the “quorum quenching” lactonase enzymes AiiA and AttM, frequently found in Gram-positive and Gram-negative bacteria, respectively [
33]. Quorum quenching enzymes are capable of hydrolyzing the lactone ring of AHL molecules.
Another target related to β-lactam antibiotics is a protein that contains an ErfK/YbiS/YcfS/YnhG domain (also known as YkuD domain) which codes for a transpeptidase that functions as an alternative pathway for peptidoglycan cross-linking. This pathway provides resistance to β-lactam antibiotics because it functionally replaces the sensitive penicillin-binding protein (PBP).
We found four transport-related putative targets. The first two are the divergently organized genes encoding a TonB-dependent receptor and an ABC transporter (of the ABC_DR_subfamily_A). The TonB-dependent receptor is involved in the specific transport of substances across the membrane, of which iron-carrying siderophores are the best characterized ones (Conserved domain database (CD) number: cd01347). Nevertheless, a role in QS has been demonstrated for TonB in
Pseudomonas aeruginosa. There, TonB mutants show an impairment in AHL production independently of its iron transport activity [
34]. On the other hand, the extensively studied ABC transporter protein family is involved in similar processes including siderophore, drug, bacteriocin, glycoconjugate and peptide efflux (CD number: cd03230). A third transport-related protein is another ABC transporter present in the set is a “toluene tolerance protein” (Ttg2 family). However, it is unlikely that toluene itself is the substrate for this transporter (CD number: cl01074). Finally, the MarC-family protein spans the membrane several times and is also thought to convey multiple antibiotic resistance, however its precise activity is unknown (CD number: cl000919).
Five genes related to RNA are also predicted targets. The first one is a TrmH-family protein, which is known to methylate rRNA. rRNA methylation in bacteria strongly correlates with resistance to ribosome-targeted antibiotics [
35]. A second activity related with RNA modification is a tRNA psudouridine synthase (TruB in
E. coli). TruB catalyzes the isomerization of specific uridines in tRNA. The third RNA-related gene is the aspartyl-tRNA synthase. Finally, the RibonucleaseT (RNAseT) protein in
E. coli RNAseT is involved in tRNA turnover. It has a 3′-5′exonuclease activity, also related with DNA polymerase III. RNAseT is only found in gamma-proteobacteria (CD number: cd06134).
Several proteins with diverse activities where found to be part of this putative regulon. First, a conjugal transfer protein (TrbF), thought to be part of the pilus required for the transfer of DNA. DNA transfer is a major QS-regulated phenotype in
Agrobacterium tumefaciens [
36]. The rest of the genes are a NADPH-dependent FMN reductase, a Prophage-related conserved protein and several hypothetical proteins.
Finally, two glycosyl-transferase activities were found among the putative regulon. These activities are usually related to polysaccharide biosynthesis. The two genes belong to different groups. The “Group 1” glycosyl-transferase protein transfer activated nucleotide-sugar molecules to several substrates, including protein, lipids and other sugar residues. The “Group 2” glycosyl transferase has the same activity but structurally it belongs to a different family (CD number: cd04186). It is possible that these proteins play a role in the biosynthesis of the exopolymeric substances (EPS) or the lipopolysaccharide. In addition,
zwf a gene encoding for glucose 6-phosphate-1-dehydrogenase (G6PDH) has been also identified as part of this regulon. G6PDH is part of the pentose phosphate pathway and is directly related to the cellular pool of glucose 6-phosphate that is involved in the biosynthesis of EPS precursors UDP-glucose and UDP galactose [
37]. On the other hand, mutation in
zwf leads to an approximately 90% reduction in alginate, one of the EPS produced by
Pseudomonas aeruginosa [
38]. It is well established in several bacterial models that EPS synthesis is regulated by quorum sensing [
8]. On the other hand, several studies have established that
At. ferrooxidans is capable of synthesizing the EPS precursors UDP-glucose and UDP-galactose [
39] and EPS is involved in the adhesion to solid surfaces by this microorganism [
40,
41]. Moreover, qRT-experiments indicated recently that transcription levels of
zwf and
afeI are increased in the presence of a tetrazolic AHL-analogue [
42,
43] suggesting that QS and EPS synthesis might be also connected in this microorganism.